Survival of the fittest 9

Discussion of computer chess matches and engine tournaments.

Moderator: Ras

enhorning
Posts: 342
Joined: Wed Jan 05, 2011 10:05 pm

Re: Survival of the fittest 9

Post by enhorning »

Tom Likens wrote:If you still have it please send me as much of the game information as you can find for the Djinn forfeiture, (I'm assuming this was a loss on time). Djinn may not be the strongest engine in existence, but I do like it to be stable and as bug-free as possible, so I take these bug reports seriously.

Also, if you wouldn't mind please send me the name of the exact Djinn executable you used as well as the particulars of the computer you ran it on, (i.e. memory, any configuration files, was ponder on or off, the name of the GUI you ran the tournament under etc., basically, anything that you think would help).

thanks,
--tom
Hi,

I'm afraid I don't have much info beyond my setup and the PGN - I have logging turned off.

Games are played under Arena 3, with a 6+4 time control, ponder off - because of the increments, all time losses I've seen has been because an engine stops responding - and it seems so here too, as Djinn had a forced mate on the board, and had moved instantly the previous few moves. Looking at the game, I wonder if the problem might be TB related, as the position had just been reduced from 5 to 4 pieces when Djinn stopped moving (I have all TBs up through 5 men installed on harddrive.)

The exact Djinn file was djinn992_x64_popcnt.exe . My computer is a 6-core i-970, running Windows 7 professional, and with 6 GB RAM.

PGN of the game:

Code: Select all

[Event "Survival 9"]
[Site "ENHORNING"]
[Date "2013.12.24"]
[Round "2"]
[White "Deuterium 13.1.31.113 [1c]"]
[Black "Djinn 0.992 [1c]"]
[Result "1/2-1/2"]
[BlackElo "2347"]
[ECO "A07"]
[Opening "Reti Opening"]
[Time "10:14:05"]
[Variation "KIA, 2.g3 Nc6 3.Bg2 e5 4.d3 Nf6 5.O-O Be7"]
[WhiteElo "2423"]
[TimeControl "360+4"]
[Termination "time forfeit"]
[PlyCount "173"]
[WhiteType "program"]
[BlackType "program"]

1. h3 e5 2. g3 Nc6 3. d3 {-0.64/16 18} d5 {+0.66/17 26} 4. Nf3 {-0.61/16
14} Nf6 {+0.61/19 24} 5. Bg2 {-0.60/17 16} Be7 {+0.61/19 23} 6. O-O
{-0.57/16 14} O-O {+0.68/16 22} 7. Nbd2 {-0.51/15 16} Bf5 {+0.60/15 19} 8.
c4 {-0.51/16 14} d4 {+0.55/17 19} 9. Nb3 {-0.42/16 14} h6 {+0.51/17 18} 10.
g4 {-0.49/16 14} Bh7 {+0.49/16 16} 11. Ne1 {-0.22/16 13} Qd6 {+0.52/15 16}
12. f4 {-0.39/14 10} Nd7 {+0.37/17 19} 13. fxe5 {-0.30/15 12} Ndxe5
{+0.55/16 14} 14. Bf4 {-0.12/16 11} a5 {+0.32/14 12} 15. e3 {0.00/15 12}
dxe3 {+1.29/14 16} 16. Bxc6 {-0.60/15 11} bxc6 {+1.14/17 21} 17. d4
{-0.82/17 11} f6 {+1.24/16 9} 18. Nf3 {-1.02/16 10} a4 {+0.70/14 15} 19.
Nc5 {-1.47/16 10} Rfb8 {+0.92/16 16} 20. Nxa4 {-1.51/16 10} Nxf3+ {+0.43/15
8} 21. Rxf3 {-0.27/13 1} e2 {+0.36/17 10} 22. Qd2 {-0.25/17 10} Qb4
{+0.71/15 10} 23. b3 {0.00/18 7} g5 {+0.56/18 9} 24. Be3 {-0.05/18 9} Rxa4
{+0.30/17 10} 25. bxa4 {+0.32/14 1} Qxc4 {+0.06/17 6} 26. Bf2 {+0.38/16 9}
Bb4 {+0.99/15 8} 27. Qb2 {+0.24/17 9} Re8 {+0.81/17 8} 28. Rc1 {+0.18/19 8}
Qxc1+ {+0.86/19 8} 29. Qxc1 {+0.35/20 1} e1=Q+ {+0.91/25 5} 30. Bxe1
{+0.16/21 8} Rxe1+ {+0.92/25 5} 31. Qxe1 {+0.30/20 1} Bxe1 {+0.78/25 5} 32.
Rxf6 {+0.24/21 8} Bc3 {+0.78/25 5} 33. Rxc6 {+0.06/22 8} Bxd4+ {+0.83/24 5}
34. Kf1 {+0.07/22 8} Bb6 {+0.83/23 5} 35. Rxh6 {0.00/22 8} Bd3+ {+0.86/22
5} 36. Kg2 {0.00/20 8} Kg7 {+0.86/19 5} 37. Re6 {-0.16/21 8} Kf7 {+0.90/20
5} 38. Rh6 {-0.18/20 8} Bc2 {+0.92/19 5} 39. Kf3 {-0.21/19 8} Bxa4
{+1.04/20 4} 40. Ke4 {-0.18/18 6} Bc2+ {+1.01/19 5} 41. Ke5 {-0.12/18 7}
Be3 {+1.05/20 5} 42. Rf6+ {-0.21/18 7} Ke7 {+1.07/20 5} 43. Re6+ {-0.24/18
7} Kd7 {+1.07/18 5} 44. Rf6 {-0.25/18 7} Bf4+ {+1.17/20 4} 45. Kd5
{-0.15/19 7} Bd6 {+1.15/22 5} 46. Rf7+ {-0.20/20 7} Be7 {+1.11/22 5} 47.
Rf2 {-0.18/20 7} Ba4 {+1.09/19 5} 48. Ke5 {-0.18/20 7} Bb5 {+1.09/19 5} 49.
Rd2+ {-0.17/19 7} Kc6 {+1.08/19 5} 50. Kf5 {-0.07/18 7} Kb6 {+1.04/17 4}
51. Rd1 {0.00/20 6} Ba4 {+1.04/19 4} 52. Rd2 {0.00/20 6} Kc6 {+0.83/18 4}
53. Rd4 {0.00/20 6} Bc2+ {+0.98/19 4} 54. Ke5 {0.00/20 6} Bh7 {+0.93/21 5}
55. Rd1 {-0.18/17 6} Bd6+ {+1.09/18 5} 56. Kf6 {0.00/20 6} Bf4 {+0.60/20 6}
57. Rf1 {0.00/21 6} Bd3 {+0.66/21 4} 58. Rf3 {0.00/21 6} Be4 {+0.91/21 6}
59. Rf2 {0.00/21 6} Kd6 {+1.50/17 4} 60. h4 {-0.14/20 5} Be5+ {+1.84/17 7}
61. Kxg5 {-0.97/20 6} c5 {+3.08/19 4} 62. Rd2+ {-0.79/19 6} Ke6 {+3.40/25
16} 63. Re2 {-1.86/20 6} Kd5 {+3.50/24 3} 64. a4 {-1.69/19 6} c4 {+3.93/19
8} 65. a5 {-2.36/19 6} c3 {+4.25/20 7} 66. a6 {-2.68/17 5} Bd4 {+4.64/21 6}
67. Kf4 {-2.97/16 5} c2 {+5.51/18 5} 68. Re1 {-3.26/16 5} Bc3 {+5.88/19 3}
69. Rf1 {-3.59/16 5} c1=Q+ {+6.22/19 5} 70. Rxc1 {-3.51/14 5} Bd2+
{+6.36/20 3} 71. Kg3 {-0.89/3 0} Bxc1 {+6.36/20 3} 72. Kf2 {-3.97/12 5} Ke5
{+6.54/18 5} 73. Kg3 {-4.56/12 5} Be3 {+6.58/15 4} 74. h5 {-5.27/12 5} Kf6
{+6.83/16 3} 75. Kh3 {-5.69/12 5} Kg5 {+7.62/15 3} 76. Kg3 {-6.10/12 5} Bd4
{+7.80/16 3} 77. h6 {-7.57/12 5} Ba7 {+M26/16 2} 78. h7 {-7.82/13 5} Bxh7
{+M21/2 0} 79. Kf3 {-8.07/13 5} Bd3 {+M13/2 0} 80. Kg3 {-8.12/13 5} Bxa6
{+M12/2 0} 81. Kf3 {-8.49/13 5} Bd3 {+M12/2 0} 82. Kg3 {-8.28/13 5} Be4
{+M11/2 0} 83. Kh3 {-8.49/13 5} Bf2 {+M10/2 0} 84. Kh2 {-8.57/13 5} Kf4
{+M9/2 0} 85. g5 {-8.44/12 5} Kf3 {+M8/2 0} 86. g6 {-M6/12 5} Bxg6 {+M7/2
0} 87. Kh3 {-M5/14 4} ... {Black forfeits on time} 1/2-1/2
Djinn's configuration file:

Code: Select all

#-*-djinn-cfg-*-
#----------------------------------------------------------
# Djinn chess program configuration file.  All lines 
# that begin with a # are considered comments. The
# lines can be reordered by the identifier fields
# should not change.
#----------------------------------------------------------

# Generate a log file (true/false)
create-log= false

# Directory to store all log files.  This can also be set by setting 
# Environment variable equivalent: DJINN_LOG_PATH
log-path ="C:\Program Files (x86)\Arena\Engines\Djinn\logs"

# Load the opening book if found (true/false)
use-opening-book = false

# Opening book to use.  
# Environment variable equivalent: DJINN_OPENING_BOOK
opening-book= C:\Program Files (x86)\Arena\Engines\Djinn\djdjinn.bin

#-----------------------------------------------------------------------
# This determines the point-of-view of the score being displayed.   
#-----------------------------------------------------------------------
#   WHITE - Always show the score from white's point-of-view (default)
#   BLACK - Or always from black's point-of-view                      
#   ENGINE- Use the engine's perspective for all scores               
#   PLAYER- Use the player's perspective for all scores displayed     
#-----------------------------------------------------------------------
pov-display = white

#---------------------------------------------------------------------
# Opening book style.  Presently there are four different style types:
#---------------------------------------------------------------------
# SOLID (default) : The engine selects a "solid" move (whatever 
#                   that means :)
# FREQ            : The engine *always* picks the move that appears
#                   most often.
# RANDOM          : Selects a completely random move (very wild)
# FIRST           : Selects the first move listed for the position
#---------------------------------------------------------------------
opening-style = SOLID

# Size of main hash table.  The size can be in megabyte (MB), 
# kilobytes (kB) or raw size (e.g. 23405056).  The value will get
# truncated to the next smaller size down.  The default = 24MB
hash = 256MB

# Same deal for the pawn hash table (default=8M).
pawn-hash = 64MB

# Set the size of the evaluation hash table (default=16M)
eval-hash = 64MB

# Load the endgame tablebases if found (true/false).
use-tb = true

# Path to the Nalimov endgame tablebase (if present)
# Environment variable equivalent: DJINN_TB_PATH
tb-path = C:\NalimovTB

# This sets the size of the Endgame tablebase RAM cache (default=8M).
tb-cache= 64M

#------------------------------------------------------------------------
# PROGRAMMING OPTIONS - These options affect the way the engine plays,
# which means you should probably leave them alone unless you have some
# idea of what you are doing, (of course, that didn't stop me! :)
# As the program's public life matures this section will grow.
#-----------------------------------------------------------------------

# This is used to pull the *enemy* pieces towards the less safe king.
# Values can range from -2.00 to 2.00 (a negative value will *REPEL* 
# the enemy pieces from the weaker king).  125 = 1.25 internally.
# (default=125)
king-safety-scalar= 125

# Resignation cutoff score in terms of pawns (3 = three pawns etc.)
# Values can range from 3 to 50. Default = 7
resignation-cutoff=50
Tom Likens
Posts: 303
Joined: Sat Apr 28, 2012 6:18 pm
Location: Austin, TX

Re: Survival of the fittest 9

Post by Tom Likens »

Thanks Ola,

This is good information. I'll start digging and see if I can reproduce the problem. If I can reproduce it, it should be easy to fix. Reproducibility is the real challenge. It's amazing how even with the best of intentions and literally hundreds of thousands of games things can still slip through. I can't tell you how many assertions Djinn has in debug mode. That's why testers are so valuable.

regards,
--tom
enhorning
Posts: 342
Joined: Wed Jan 05, 2011 10:05 pm

Re: Survival of the fittest 9

Post by enhorning »

SzG wrote:Daydreamer is a notorious forfeiter in my system. I have 64-bit Vista, Q6600 CPU. What is yours? It might be OS dependent.
64-bit Windows 7 Professional, hexacore i7-970.